Softening effect refers to the inherent property of porous media materials of having their strength decrease over time under water saturation compared with that under dry condition. Typical porous media materials, such as ceramsite and ceramsite-based concrete, exhibit a softening effect. This study investigated the softening effect of shale ceramsite (SC) and different types of lightweight aggregate concrete (LWACs). Mix ratios of LC30 all-lightweight shale ceramsite concrete (also known as all-lightweight concrete, ALWC) whose SC and shale pottery (SP) were selectively or simultaneously replaced with limestone gravel (LSG) and river sand (RS) with the same volume fraction, respectively, were prepared for comparison with gravel lightweight concrete (GLWC), sand lightweight concrete (SLWC) and hybrid aggregate lightweight concrete (HALWC) were formed. The four kinds of LWACs (ALWC, GLWC, SLWC and HALWC) were cast into their corresponding specification and size specimens and demoulded after indoor curing for 24 h, with their target ages (7–180 d) maintained under standard and hydrostatic curing. Then, the strength and initial elastic modulus under uniaxial stress, ultimate compressive strength and stress–strain curves under triaxial stress of the specimens were tested to analyse the influence of aggregate types and curing methods on the mechanical properties and softening effect of LWACs. According to the test results, the cube compression, prism compression, splitting tensile, conventional triaxial compression and elastic modulus decreased after 180 d of hydrostatic curing; the experimental age of the conventional triaxial compression test was after 90 d. ALWC demonstrated the most evident decrease, followed by SLWC, HALWC and GLWC. The trend indicates that normal aggregates can significantly improve the softening effect of LWACs. The failure modes of the triaxial compression specimens changed from splitting failure to shear failure, and a plastic platform phenomenon was apparent on the stress–strain curves, with their descending section appearing gentler than those in the uniaxial compression test with the increase in confining pressure. The relationship between ultimate compressive strength and confining pressure of the LWACs can be accurately described by the Mohr–Coulomb single-parameter failure criterion, Hsieh–Ting–Chen four-parameter failure criterion and Willam–Warnke five-parameter failure criterion. However, the three failure criteria demonstrated application limitations based on meridian equation and deviatoric plane analysis. The extension trend of the compression meridian of the LWACs after 90 d of hydrostatic curing was generally the same, and the types of aggregates exerted a minimal effect on the meridian and deviatoric plane limit trace. The softening effect of LWACs was mainly caused by the softening effect of the SC itself and the influence of pore water on the matrix concrete.